Safety overview: GPT-6 Astra

Why it matters

Why it matters: GPT-6 Astra's "Critical" cybersecurity rating signifies a new benchmark in AI safety preparedness, particularly for advanced models. This level suggests robust defenses against misuse and adversarial attacks. Competitors like Google Gemini and Anthropic's Claude are also heavily invested in AI safety, but OpenAI's explicit framework and rating provide a concrete measure. Future releases will likely see other labs adopt similar public-facing safety metrics.